Impact of fracture‐prone implantable cardioverter defibrillator leads on long‐term patient mortality

Abstract Background The long‐term relationship between fracture‐prone implantable cardioverter‐defibrillator (ICD) leads and poor prognosis remains unclear in Japanese patients. Methods We conducted a retrospective review of the records of 445 patients who underwent implantation of advisory/Linox leads (Sprint Fidelis, 118; Riata, nine; Isoline, 10; Linox S/SD, 45) and non‐advisory leads (Endotak Reliance, 33; Durata, 199; Sprint non‐Fidelis, 31) between January 2005 and June 2012 at our hospital. The primary outcomes were all‐cause mortality and ICD lead failure. The secondary outcomes were cardiovascular mortality, heart failure (HF) hospitalization, and the composite outcome of cardiovascular mortality and HF hospitalization. Results During the follow‐up period (median, 8.6 [4.1–12.0] years), there were 152 deaths: 61 (34%) in patients with advisory/Linox leads and 91 (35%) in those with non‐advisory leads. There were 32 ICD lead failures: 27 (15%) in patients with advisory/Linox leads and five (2%) in those with non‐advisory leads. Multivariate analysis for ICD lead failure demonstrated that the advisory/Linox leads had a 6.65‐fold significantly greater risk of ICD lead failure than non‐advisory leads. Congenital heart disease (hazard ratio 2.51; 95% confidence interval 1.08–5.83; p = .03) could also independently predict ICD lead failure. Multivariate analysis for all‐cause mortality demonstrated no significant association between advisory/Linox leads and all‐cause mortality. Conclusions Patients who have implanted fracture‐prone ICD leads should be carefully followed up for ICD lead failure. However, these patients have a long‐term survival rate comparable with that of patients with non‐advisory ICD leads in Japanese patients.


| INTRODUC TI ON
The role of implantable cardioverter-defibrillators (ICDs) in reducing mortality from life-threatening arrhythmias in high-risk patients has been established. 1,2 The failure of high-voltage leads can compromise ICD function, causing inappropriate ICD shocks, defibrillation failure, fatal proarrhythmias, and even death. 3,4 The Sprint Clamart, France) were withdrawn from the market as a result of an increased rate of lead fracture compared with leads from other manufacturers. [5][6][7] Linox S/SD leads (Biotronik, Berlin, Germany) also reportedly had unacceptably high rates of lead failure. [8][9][10] Several previous studies have been published on the relationship between these fracture-prone ICD leads and mortality compared with nonadvisory leads. 11,12 However, the follow-up periods were relatively short (3-4 years). 11,12 Furthermore, these previous studies were almost entirely from Europe or North America. Data are scarce on the relationship between fracture-prone ICD leads and mortality in the Asian population compared with non-advisory leads. Therefore, this study aimed to investigate the long-term relationship between fracture-prone ICD leads and poor prognosis (all-cause mortality and cardiovascular adverse events) in Japanese patients.  All study ICD leads were implanted by experienced cardiologists specializing in electrophysiology. ICD leads were implanted via a left-or right-sided cephalic vein by cut-down, axillary, or subclavian veins using standard puncture techniques. ICD leads were positioned in the right ventricular apex or the ventricular septum. After lead implantation, the R-wave amplitude, lead impedance, and pacing threshold were evaluated in all patients. After testing, they were connected to the ICD pulse generator, and the incision was closed.

| ME THODS
After ICD implantation, patients were followed up every 6 months using a remote monitoring system (RMS) or through ambulatory follow-ups at our hospital. The lead integrity alert (LIA) algorithm was used to detect early lead fractures, as described in a previous report. 13 In accordance with the previously reported ICD lead survival rates, 5-10,14-16 patients were categorized into two groups: patients implanted with fracture-prone ICD leads (Sprint Fidelis, Riata, Isoline, or Linox leads) and those implanted with non-advisory leads (Durata, Endotak Reliance, or Sprint non-Fidelis leads). Linox leads have been reported to have comparable lead survival rates to those of advisory leads. [8][9][10] Therefore, patients implanted with Linox leads were grouped with those with advisory leads.
ICD lead failure was defined as a lead removed from service because of its inability to meet its performance specifications or otherwise not performing as intended. In line with previously established definitions, 11,14,15 a lead had failed if it met one or more of the following criteria: (i) a sudden rise in long-term pacing or highvoltage impedance, (ii) electrical noise artifacts manifested by nonphysiological signals on the electrogram or by device diagnostics (e.g., non-physiological short intervals and/or recurrent non-sustained ventricular tachycardia with intervals <220 ms), and (iii) a failure to sense R-waves or ineffective electrical therapy caused by an apparent structural lead defect. Cases with pseudofractures, because of the header or connector problems, were excluded at the time of replacement. Functional abnormalities, including exit block and physiological oversensing with an electrically intact lead, were excluded from the definition of ICD lead failure. Lead dislodgement was not considered an ICD lead failure. A lead-related death was defined as a sudden or unexpected death accompanied by evidence of an ICD lead failure. A group of 100%-paced patients were categorized by complete heart block in the absence of a stable escape rhythm or 100% pacing in patients without cardiac resynchronization therapy (CRT). The primary outcomes of the study were all-cause mortality and ICD lead failure. The secondary outcomes were cardiovascular mortality, heart failure (HF) hospitalization, and the composite outcome of cardiovascular mortality and HF hospitalization.
According to the distribution and dispersion, continuous variables were expressed as means ± standard deviations or as medians with interquartile ranges (IQR

| RE SULTS
A total of 445 patients (Sprint Fidelis, n = 118; Riata, n = 9; Isoline n = 10; Linox, n = 45; Durata, n = 199; Endotak Reliance n = 33; Sprint non-Fidelis n = 31) who underwent implantation of an ICD lead were included in the study. The baseline characteristics of patients with advisory/Linox leads and non-advisory leads are shown in Table 1. The median follow-up period, until the last follow-up date, in patients with advisory/Linox leads was significantly longer than that in patients with non-advisory leads. The prevalence of HF and screw-in lead use in patients with non-advisory leads were significantly higher than that in patients with advisory/Linox leads. The prevalence of a lead length ≥65 cm in patients with advisory/Linox leads was significantly higher than that in patients with non-advisory leads. There were significant differences in age and left ventricular ejection fraction (LVEF) between the two groups. Table S1  The fracture sign and lead-related deaths in patients with ICD lead failure for each ICD lead type are shown in Table 2. During follow-up, until ICD lead removal or revision date, we identified ICD lead failure in 20 Sprint Fidelis leads (17%), 1 Riata lead (11%), 1 Isoline lead (10%), 5 Linox leads (11%), 4 Durata leads (3%), and 1 Sprint non-Fidelis lead (3%). The median interval from implantation to ICD lead failure was 6.0, 12.4, 9.3, 8.3, 4.7, and 9.5 years with Sprint Fidelis, Riata, Isoline, Linox, Durata, and Sprint non-Fidelis leads, respectively. In 22 leads, a sudden rise in long-term pacing or high-voltage impedance was found. In 16 leads, electrical noise artifacts were found, and 12 patients experienced inappropriate shocks because of the electrical noise artifacts. The LIA was activated in 20 patients and prevented inappropriate shocks in 13 of them. No lead-related death occurred in any patient with ICD lead failure. An externalized conductor was found in a fractured Riata lead. Table 3 shows the hazard ratio (HR) of ICD lead failure, until the ICD lead removal or revision date, based on the univariate analysis. Advisory/Linox, Sprint Fidelis, and Linox leads were significantly associated with increased ICD lead failure. Other devices and patient factors associated with increased ICD lead failure in univariate analysis were congenital heart disease and the total number of leads at implantation. Multivariate analysis was performed according to the results of the univariate analysis. Advisory/Linox leads had a 6.65-fold higher risk of ICD lead failure than non-advisory leads (HR, 6.65; 95% confidence interval [CI], 2.55-17.4; p = .0001). Moreover, congenital heart disease could also independently predict ICD lead failure (HR, 2.51; 95% CI, 1.08-5.83; p = .03). Figure 1 shows the Kaplan-Meier curves of the ICD lead survival rates for advisory/Linox leads and non-advisory leads. During the follow-up period, until the ICD lead removal or revision date, ICD lead failure occurred more frequently in patients with advisory/ Linox leads than in patients with non-advisory leads (p < .0001). The 10-year ICD lead survival rates were 80% and 97% for advisory/ Linox leads and non-advisory leads, respectively. Figure S1 shows the Kaplan-Meier curves of ICD lead outcome-free survival rates for each ICD lead type. There were significant differences in ICD lead survival rates among the different ICD lead types (p = .0002).
In particular, Sprint Fidelis leads fractured more frequently than all other non-advisory leads. Table 4 shows the HR of all-cause mortality based on the univariate and multivariate analyses. In the univariate analyses, advisory/Linox leads and an ICD lead type were not significantly associated with all-cause mortality. Conversely, age, CRT, HF, ischemic cardiomyopathy, congenital heart disease, LVEF, atrial fibrillation (AF), elevated creatinine, ICD appropriate shock, and ICD inappropriate shock were associated with increased all-cause mortality in the univariate analyses. Based on the results of univariate analyses, multivariate analyses were performed in two models with variables that were considered to be related to allcause mortality. Model 1 considered whether fracture-prone ICD leads were implanted, while Model 2 considered the influence of fracture-prone ICD lead type (Sprint Fidelis, Riata, Isoline, or Linox) on all-cause mortality. The advisory/Linox leads and all fracture-prone ICD lead types were not significantly associated with increased all-cause mortality compared with non-advisory leads. However, advanced age, elevated creatinine levels, HF, and AF were independent predictors of all-cause mortality. In the multivariate analyses for all-cause mortality, including each reason for ICD inappropriate shock, the reasons for ICD inappropriate shock were not significantly associated with all-cause mortality (Table S2).  leads. There were no significant differences between the two groups in the incidence of all-cause mortality, the outcome of cardiovascular mortality, HF hospitalization, and the composite outcome of cardiovascular mortality and HF hospitalization (Figures 2A-D). Figure S2 shows

| DISCUSS ION
Our study has four main findings. First, the multivariate analysis for ICD lead failure found that advisory/Linox leads and congenital heart disease were significantly associated with ICD lead failure in the long term. Second, no lead-related death occurred in any patient with confirmed ICD lead failure. Third, the multivariate analysis for all-cause mortality found that using fracture-prone ICD leads did not  3,4,17 In particular, patients who were implanted with advisory leads could be exposed to a more frequent risk of ICD lead failure than those who received non-advisory leads. [5][6][7][14][15][16][18][19][20] With respect to Linox leads, not advisory leads, several previous studies have suggested that the unacceptably high rate of lead failure contradicts the self-reported data from the manufacturer. [8][9][10] Nevertheless, few studies have evaluated whether using fractureprone ICD leads to increased mortality compared with using nonadvisory leads. 11,12 Some previous studies have shown that Sprint Fidelis leads did not increase all-cause mortality in the relatively short period of 2.9-3.2 years, compared with Sprint Quatro leads. 11,12 Furthermore, these studies were limited to one type of advisory lead. 11 (Figures 2B-D). The median follow-up periods in TA B L E 2 Fracture sign and ICD lead-related death in patients with ICD lead failure.   patients implanted with Sprint Fidelis, Riata, Isoline, and Linox leads were 8.5, 6.0, 11.5, and 11.5 years, respectively Table S1. These follow-up periods were longer than those in the previous studies that investigated the relationship between fracture-prone ICD leads and mortality. 11,12 Furthermore, patients continued follow-up after ICD lead removal or revision. Therefore, the design of the current study allowed the consideration of the impact of ICD inappropriate shocks and ICD lead extraction, which were often accompanied by ICD lead failure, on prognosis.
In the multivariate analysis, having an advanced age, a decline in renal function, HF, and AF were independent risk factors for allcause mortality (Table 4). Several previous studies have shown that patients implanted with ICD who have advanced age, or AF or HF, have a poorer prognosis than those without these risk factors. 23,24 Hager et al. showed that chronic kidney disease was an independent predictor of mortality in patients who underwent ICD implantation for primary prevention. 25 These findings support our results.  [18][19][20] Moreover, some earlier studies have shown that Riata, Isoline, and Linox leads had unacceptably higher rates of lead failure than other leads. [5][6][7]16,18 Similarly, the present study demonstrated that advisory/Linox leads had a 6.65-fold greater risk of ICD lead failure than non-advisory leads, after adjusting for cofounders that were considered to be related to ICD lead failure (Table 3). Moreover, Sprint Fidelis and Linox leads had a significantly greater risk of ICD lead failure than that non-advisory leads ( Table 3). In contrast, Riata and Isoline leads did not have a significantly greater risk of ICD lead failure in the present study; however, this may have been because of the smaller number of patients implanted with Riata leads (n = 9) and Isoline leads (n = 10).
The small patient number may have resulted in Riata and Isoline leads not demonstrating a significantly greater risk of ICD lead failure than non-advisory leads.
In Durata lead failures, the frequencies of electrical noise artifacts and ICD inappropriate shocks were lower than those when F I G U R E 2 Difference in outcome-free survival rates between patients with advisory/Linox leads and those with non-advisory leads. Kaplan-Meier curves showing the differences in the cumulative rates of the following clinical outcomes until the last follow-up date: (A) All-cause mortality, (B) Composite outcome, (C) Cardiovascular mortality, (D) Heart failure hospitalization. ICD, implantable cardioverterdefibrillator.
using the Sprint Fidelis or Linox leads (Table 2). Although no direct comparison between the Durata leads and the other two leads was made, single-coil Durata models are less likely to develop internal insulation breaches than dual-coil Durata models. 28 Internal insulation breaches could disrupt the silicone, abrade the fluorine-based protective coating, and short-circuit conductors to each other or to the shock coils. 28 The internal insulation breaches may present as electrical noise artifacts with ICD inappropriate shocks. 28 In the present study, the models of fractured Durata leads were the Durata 7120 (n = 1), 7122 (n = 2), and 7122Q (n = 1). Therefore, we were unable to adjudicate the probability of death because of the ICD lead failure in some patients. Finally, the study was a retrospective analysis from a single center.

| CON CLUS IONS
Using advisory/Linox leads and having congenital heart disease independently predicted ICD lead failure. However, advisory/Linox leads were not associated with increased all-cause mortality in extended long-term follow-up in Japanese patients. Patients who were implanted with fracture-prone ICD leads should be carefully followed up for ICD lead failure. However, such patients have comparable long-term survival rates as those with non-advisory ICD leads.

ACK N OWLED G M ENTS
The authors would like to thank Yasuto Sato, an instructor in the Department of Public Health at Tokyo Women's Medical University, for his advice on statistics. The authors would also like to thank Editage (www.edita ge.com) for English language editing.

CO N FLI C T O F I NTER E S T S TATEM ENT
None.

I N FO R M ED CO N S ENT
As a result of the retrospective nature of the study, informed consent was not necessary, and the opt-out method was used through the hospital website.